Real-Time Facial Recognition System

A pipeline to move facial recognition out of the lab and into the "messy" real world, focusing on maintaining high accuracy across different lighting and moving subjects.

• Core Stack: Integrated OpenCV for real-time frame processing and FaceNet (PyTorch) to generate high-dimensional embeddings for near-instant identity matching.
• Affective Computing: Added an emotion analysis library to categorize facial expressions.
• Environmental Robustness: Developed a preprocessing stage to normalize lighting and orientation.
• Heuristic Tuning: sensitive enough for quick detection but strict enough to eliminate false positives in live video.